Skip to content

Latest commit

 

History

History
47 lines (35 loc) · 2.67 KB

File metadata and controls

47 lines (35 loc) · 2.67 KB

Roadmap

The toolkit's strategy is simple: build every surface on the one shared engine. New interfaces are thin adapters, so they inherit the same read-only, offline-first, deterministic behaviour for free. See Architecture.

Available now

  • CLI (devops-ai) — analyze, explain, validate, list, serve, version. See CLI guide.
  • Python SDKAnalysisEngine and the public models. See SDK guide.
  • REST API — FastAPI with Swagger/ReDoc. See REST API guide.
  • Extensibility — YAML signatures and pluggable AI providers via register_provider(). See Plugin guide.
  • Providers — Anthropic, OpenAI, Gemini, Ollama, plus the offline default.

Planned

These are design directions, not commitments. Each reuses the engine — none reimplements analysis.

Surface Idea Design note
Web UI Browser front-end over the REST API for paste-and-analyze + browsing web-ui.md
VS Code extension Analyze the active file / selection / terminal output inline vscode-extension.md
GitHub Action Analyze CI logs and manifests in pipelines; annotate PRs github-action.md
MCP server Expose analyze/explain/validate as tools to AI agents mcp-server.md
Desktop app Offline, local desktop wrapper for operators

Ongoing

  • Broader knowledge base — more signatures across the supported technologies and new ones. Contributions welcome; it's just YAML. See Knowledge base.
  • Smarter detection and ranking — improved technology/source-kind detection and scoring.
  • More providers — additional adapters as the ecosystem evolves.

Principles that won't change

  • Read-only. No surface will ever execute commands or mutate infrastructure. See Security.
  • Offline-first & deterministic. The core never requires an API key; LLMs stay optional.
  • One engine. Every interface is a thin adapter over AnalysisEngine.
  • Additive contracts. AnalysisResult grows by addition so interfaces never drift.

Get involved